Data from 4771 patients with tumor diameters < or = 10 mm were analyzed. Results of surgery and pathoanatomical examinations indicated that nodal status was related to diameter, but not to number of nodes removed. More axillary metastases were found in group T1b tumors than in T1a. In 8% of tumors, at least 4 positive nodes were identified. Mean number of positive nodes was related to number of nodes removed, and when 10 or more nodes were removed a significantly lower axillary recurrence rate and better recurrence-free survival were demonstrated, confirming that axillary surgery has two goals: staging and regional disease control. Age, receptor status, grade and histological type, but not tumor location, were related to prognosis. In accordance with the classical prognostic factors, it was not possible to define a patient group where axillary surgery was superfluous. We conclude that proper staging and regional control renders a full axillary level I-II dissection necessary.